Re: open(2) support for sockets

[William Ahern <william_(_at_)_25thandClement_(_dot_)_com>]

i've patched some source files in busybox and the ash shell which it
includes. simple changes to open() and fopen() in those sources seemed
to reflect perfectly in the i/o redirection tests i did. before i start
begin to patch libc privately, does anybody know of some caveats i should
be on the look-out for?

fwiw, since domain sockets exist within the regular fs namespace, i'd
imagine that the burden of proof should rest w/ those averse to the idea,
not w/ the proponents. not being to read/write an object on the filesystem
seems like more of a bug than a missed feature.

- Bill